Re: let binding with c

hello,

i'm using bigloo4.2c and have had a similar issue last year,i'm using bigloo with the java option enabled to generate JVM byte code and i noticed that 
when creating string variables with nested define in function ,the code :

(define (ResultatMesuresFbeta bstr_identificateur bstr_objet)
  
  (display-var-nl "BiglooCode.scm :: ResultatMesuresFbeta :: bstr_identificateur = " bstr_identificateur)
  (display-var-nl "BiglooCode.scm :: ResultatMesuresFbeta :: bstr_objet = " bstr_objet)
    
  (define len_bstr_identificateur (byte*-length bstr_identificateur))
  (display-var-nl "BiglooCode.scm :: ResultatMesuresFbeta :: len_identificateur = " len_bstr_identificateur)
  (display-var-nl "BiglooCode.scm :: ResultatMesuresFbeta :: (string? bstr_identificateur) = " (string? bstr_identificateur))

  (define identificateur (make-string len_bstr_identificateur))
  ;;(define objet "")

  (set! identificateur  bstr_identificateur)
  ;;(set! identificateur  (string-append bstr_identificateur))

  ;;(display-var-nl "BiglooCode.scm :: ResultatMesuresFbeta :: (string-append \"\" bstr_identificateur) = "  (string-append "" bstr_identificateur))
  ;;(set! objet bstr_objet)

i had to set! the variable ,hoping it will works, finally i replaced all the nested define in function by a let* and it was ok

the code is compiled to generate byte-code running in a tomcat8 server so i did not post in bigloo list at the time it happened. 

Regards,
Damien
Le Wednesday 19 October 2016 09:33:30 Bruant, Michel, vous avez écrit :
> Hi Bigloo team,
> 
> We observe an issue with the latest bigloo-4.3  (it was already present in the previous  beta)
> I have a procedure which has a let* . In this let*, a variable is binded to a C function, something like:
> (data        (pragma::bstring "string_to_bstring_len( $1, $2 )" [....])
> Then this variable is used in some of the next bindings, and also in the body of the let*.
> But at one stage, (in the body if I remember) ; its content is wrong, corrupted.
> 
> As a workaround, we needed to change this implementation:
> - create a global variable (define data "") ; split the let* into 2 let* and add  a  (set! ...) in between to replace the previous binding.
> ... something a little bit dirty.
> I 'm not sure I can create a small test case to observe this issue; but maybe this description is enough ?
> Anyway, If you can find a fix  so that  we can remove our workaround, it will be really appreciated.
> Thanks, regards,
> Michel.
> 
> 
> 



-- 
[email protected], [email protected], UNS / OCA / CNRS